JP2022550373A - Methods and systems for determining cardiovascular parameters - Google Patents

Abstract

A system and method for determining a cardiovascular parameter comprises receiving a plethysmogram (PG) dataset; removing noise from the PG dataset; segmenting the PG dataset; and transforming the set of criteria to determine cardiovascular parameters. [Selection drawing] Fig. 3

Description

Cross-reference to related applications
[0001] This application claims the benefit of US Provisional Application No. 62/908,758, filed October 1, 2019, which is incorporated by reference in its entirety.

[0002] The present invention relates generally to the field of cardiovascular parameters, and more particularly to new and useful systems and methods in the field of cardiovascular parameters.

[0003] The human cardiovascular system includes the heart and blood vessels and plays a central role in distributing nutrients, hormones and blood cells throughout the body to maintain homeostasis and fight disease.

[0004] Measuring and monitoring the cardiovascular system has been important in human history. Many biomarkers of the cardiovascular system have been identified for prognostic and diagnostic purposes. Many of these measurements are common, including heart rate and blood pressure. The availability of others such as heart rate variability, pulse wave velocity and arteriosclerosis for medical professionals and consumers has evolved as well.

[0005] In many cases, direct measurement of these biomarkers is difficult or prohibited (eg, invasive). Intra-arterial pressure, or blood pressure, is an example that requires the use of a pressure transducer inserted within one of the major arteries for direct measurement. Because of the risks associated with arterial cannulation, non-invasive assessment of blood pressure is assessed using an inflatable cuff. These devices usually inflate an air bladder wrapped around the upper arm to temporarily block blood flow through the artery, the maximum force required to block all flow, and the force required when flow returns. and record (Korotkov method). A drawback of this method is the inability to generate real-time beat-to-beat blood pressure.

[0006] Therefore, there is a need to create new and useful methods in the field of cardiovascular detection. The present invention provides such new and useful methods.

[0007] Figure 1 is a schematic diagram of an example of the system; [0008] Figure 2 is an application flow of an embodiment of the method; [0009] Figure 2 is an application flow of one embodiment of the method; [0010] FIG. 1 is a schematic diagram of an example of processing a data set; [0011] FIG. 1 is a schematic illustration of an example of a cardiovascular manifold. [0012] FIG. 1 is a schematic diagram of an example of determining cardiovascular variants for a patient-control group and generating common cardiovascular variants. [0013] Based on the transformation from the individual cardiovascular manifold to the common cardiovascular manifold of FIG. ) to cardiovascular status. [0014] A set of criteria for each member of the control group is measured, and one or more cardiovascular parameters of the control group (e.g., _CV1 , _CV2 , ..., _CVN , e.g., blood pressure, arteriosclerosis, etc.) are measured simultaneously. and determining a transform (T) that relates a criterion to a cardiovascular parameter. FIG. [0015] FIG. 7B is a schematic diagram of an example of determining an individual's cardiovascular status based on the transformation of FIG. 7A and the individual's criteria; [0016] Possible criteria associated with the processed data set, the first derivative of the processed data set, the second derivative of the processed data set and the third derivative of the processed data set (e.g. , some or all of the circled points). [0017] FIG. 2 is a schematic diagram of an example of determining cardiovascular manifolds of a patient based on fitting corresponding to segments of a PPG data set;

[0018] The following description of preferred embodiments of the invention is not intended to limit the invention to those preferred embodiments, but rather to enable those skilled in the art to make and use the invention. intended to be

1. Overview
[0019] As shown in FIG. 2, the method 200 includes measuring S210 the arterial pressure data set, processing S220 the arterial pressure data set, analyzing S230 the arterial pressure data set, and presenting the analysis. performing S240; however, the method may include any suitable step. Processing the arterial data set includes interpolating the data set S221, filtering the data set S223, segmenting the data set S225, denoising the data set S226, and analyzing the data set. and/or any suitable step.

[0020] The method preferably functions to determine one or more cardiovascular parameters (and/or physiological parameters) of an individual. However, the method may additionally or alternatively function to predict the effect of a given activity on an individual's cardiovascular parameters and/or function in other ways. Examples of cardiovascular parameters include blood pressure, arteriosclerosis, stroke volume, heart rate, blood volume, pulse wave time, systole, pulse wave velocity, heart rate variability, blood pressure variability, drug interactions (e.g. , effects of vasodilators, vasoconstrictors, etc.), cardiovascular drift, cardiac events (e.g., thrombosis, stroke, heart attack, etc.), cardiac output, cardiac index, systemic vascular resistance, oxygen delivery, oxygen consumption, Baroreflex sensitivity, stress, sympathetic/parasympathetic tone, and/or any suitable cardiovascular parameter and/or characteristic. Examples of prescribed activities include meditation, exercise management therapy (e.g., light, moderate, vigorous, etc. exercise), medication (e.g., type, dosage, etc.), respiratory exercise, drug intake (e.g., alcohol, coffee inn, etc.) and/or any suitable activity.

[0021] The method 200 is preferably calibrated to determine an individual's cardiovascular parameters; however, the method need not be calibrated. In a first variant, the method can be calibrated independently for each individual (eg patient, user, etc.). In this variation, calibrating the method includes, for an individual, and simultaneously or substantially simultaneously (e.g., within 15 seconds, 30 seconds, 60 seconds, 2 minutes, 5 minutes, 10 minutes, etc.) ( measuring calibrated cardiovascular parameters (e.g., using a blood pressure cuff, sphygmomanometer, radial artery tonometry, arterial catheter, electrocardiogram (ECG), etc.); measuring vascular parameters. In this variation, the method further comprises generating a manifold for the individual based on criteria and cardiovascular parameters extracted from the individual's arterial pressure measurements, and from the individual's subsequent arterial pressure measurements: The extracted criteria can be mapped to cardiovascular status (eg, including values for one or more cardiovascular parameters) using the individual's manifold.

[0022] In a second variation, the method can be calibrated for any individual (eg, whether or not they have previously used the method to determine cardiovascular parameters). In a second variant, calibrating the method includes: for each individual in the control group (e.g., using a blood pressure cuff, sphygmomanometer, radial artery tonometry, arterial catheter, electrocardiogram (ECG), etc.) measuring a calibrated cardiovascular parameter, measuring the cardiovascular parameter of each individual in the control group simultaneously or substantially simultaneously (e.g., 15 seconds, 30 seconds, 60 seconds, 2 minutes) according to the method and/or substeps thereof; , within 5 minutes, 10 minutes, etc.) and determining common cardiovascular variants and/or general transformations. In a second variation, common cardiovascular manifolds and/or general transformations may be used to transform a data set associated with any individual into one or more cardiovascular parameters. For example, a general transformation may transform an individual's reference set into a cardiovascular parameter set (eg, transform a reference vector into a cardiovascular parameter vector or cardiovascular status). In a second example, a common cardiovascular manifold may be used to map a set of reference values to corresponding cardiovascular parameter values. However, the method may be calibrated based on modeling and/or simulation and/or calibrated in other ways according to physical relationships.

[0023] Calibrating the method is preferably performed while the control group individual and/or the plurality of individuals are resting (e.g., relaxing); may be performed while an individual and/or multiple individuals are active (e.g., engaging in exercise, engaging in mental gymnastics, etc.) and/or for any suitable user state . Calibrating the method preferably only needs to be performed once (e.g. for a given user, for a group of users, for all users, etc.); , monthly, yearly and/or at any suitable time. The method is preferably calibrated for the entire population, but alternatively or additionally segmented (e.g., based on age, demographics, comorbidities, biomarker combinations and/or other parameters). ) can be calibrated for population subsets. Calibrating the method preferably includes a transformation (e.g., calibration, equation, lookup table, algorithm, parameter , etc.). However, the method can be calibrated in any suitable manner.

2. advantage
[0024] Variations of the technology may provide several benefits and/or advantages.

[0025] First, variations of the present technology may allow for robust, long-term measurement of an individual's cardiovascular parameters. In certain instances, the technology may determine an individual's cardiovascular parameters months after the calibration process without recalibrating the technology.

[0026] Second, variations of the present technology may allow non-invasive determination of cardiovascular parameters for an individual. In certain examples, the technology may measure an individual's arterial pressure using a personal device.

[0027] Third, variations of the present technology may rapidly capture and determine an individual's cardiovascular parameters. In certain examples, the techniques may take advantage of computational efficiency to determine an individual's cardiovascular parameters on a near beat-to-beat basis.

[0028] Fourth, variations of the present technology provide one or more of method 200 because mobile computing devices can be transformed into biosignal detectors with high specificity in determining relevant cardiovascular parameters. Improvements may be provided to the mobile computing device itself implementing portions of. In an example, the techniques may enable assessment of cardiovascular parameters using fewer data sources (e.g., without electrocardiogram data), thus requiring fewer types of data to be processed by the computing system.

[0029] Fifth, variations of the present technology may utilize image-derived signal processing techniques to specifically determine and assess cardiovascular parameters. In certain examples of the present technology, cardiovascular parameters may enable automated facilitating therapy delivery, including adjusting medication delivery, automating environmental aspects of an individual to promote individual health. providing coordinated medical recommendations, facilitating digital communication between patients and health care providers, and/or any appropriate therapy for managing cardiovascular health Offer included.

[0030] Sixth, we found that for each individual, the relationship between the individual's baseline and various cardiovascular parameters was approximately constant; and the parameters of . This allows the use of static transformations (calibrated for the individual) to determine an individual's cardiovascular parameter values, regardless of the individual's current physiological state, and taking into account the individual's baseline values. enable We further suggest that the reference-cardiovascular parameter relationship (eg, manifold; higher-dimensional manifold) for a given individual is offset and/or a scaled version of the common manifold (eg, individual We found that all individual baseline-cardiovascular parameter relationships were approximately the same for the population as a whole, as can be expressed as the normative transformation of . This allows an individual's manifold (and/or cardiovascular status) to be determined without sampling the full range of the individual's cardiovascular status.

[0031] However, variations of the technology may provide any other suitable benefits and/or advantages.

3. system
[0032] As shown in FIG. 1, the system 100 includes one or more of a data collection module 110, a data processing module 120, a data analysis module 130, and optionally an interface device 140, and/or and any suitable components. System 100 measures an arterial pressure dataset (e.g., one or more arterial pressure waveforms) and converts the arterial pressure dataset into one or more cardiovascular parameters (e.g., in real-time, on a pulse-by-beat basis, etc.). can function as The system preferably enables one embodiment, variation or example of the method described above and/or otherwise performs, but additionally and/or alternatively, determining changes in cardiovascular parameters over time. can facilitate the performance of any suitable method, including No. 16/538,206, filed Aug. 12, 2019, entitled "Methods And Systems For Cardiovascular Disease Assessment And Management," and/or as disclosed in U.S. patent application Ser. may include one or more modules (e.g., data acquisition module, camera module, signal generation module, data processing module, data analysis module, output module, etc.), the entirety of each of the above applications being incorporated herein by this reference. .

[0033] The system 100 includes user devices (eg, smart phones, laptops, smart watches, etc.), remote computing devices (eg, clouds, servers, etc.), healthcare provider devices (eg, dedicated equipment, healthcare provider smart phones, etc.). ), and/or may include and/or be implemented in one or more of any suitable devices. In a particular example, an individual's smart phone captures multiple images and transmits the multiple images to a cloud computing server to be processed and/or analyzed.

[0034] The data collection module 110 preferably collects one or more datasets (eg, image sets, arterial pressure datasets, raw datasets, arterial pressure waveforms, photoplethysmograms (PPGs)) relating to an individual's cardiovascular parameters. ) dataset, plethysmogram dataset, etc.). The data collection module is preferably coupled to the data processing module and the data analysis module; however, the data collection module may communicate with the data processing module, may communicate with the data analysis module, and/or any It can be configured in any suitable way.

[0035] The arterial pressure data set is preferably measured on the surface of an individual's finger (e.g., detecting an arterial pressure waveform from the radial artery); , from the ulnar artery), a brachial surface (e.g., brachial artery), a chest surface (e.g., aorta), a femoral surface (e.g., femoral artery), an ankle surface (e.g., tibial artery), and/or , can be measured on any suitable body region of the individual. The arterial pressure data set is preferably a series of arterial pressure data points (e.g., 30 Hz, 60 Hz, 15 sec, 30 sec, 45 sec, 1 min, 2 min, 5 min, etc., with predetermined length of time). , 120 Hz, 240 Hz, etc.); however, the arterial pressure data set may be a single data point and/or any suitable data set. The arterial pressure data set is preferably raw data (eg, primary data); however, the arterial pressure data set can be any suitable data.

[0036] The arterial pressure data set is preferably measured using a data collection module that is in contact with the individual's body area, but may be measured using a data collection module that is not in contact with the individual's body area. . The data acquisition module preferably contacts the individual's body area with a substantially constant pressure (e.g., no more than 5%, 10%, 20%, etc. pressure changes during data acquisition), but variable or variable pressure. and/or any suitable pressure (eg, when the contact pressure is measured concurrently with the arterial pressure data set). The data collection module is preferably held in a substantially fixed orientation during data acquisition (e.g. position change less than 100 μm, 1 mm, 1 cm, 5 cm, etc.; (e.g., less than 0.1°, 1°, 2°, 5°, 10°, etc.) relative to one or more reference axes. However, the data collection module may be moved, relocated and/or turned during data acquisition (eg, to improve data quality). In some embodiments, the data collection module may include trackers and/or guides that may function to help a user maintain and/or set a predetermined contact pressure and/or orientation.

[0037] The data collection module 110 preferably includes an optical sensor (eg, a camera); , and/or may include any suitable sensor.

[0038] In variations that include an optical sensor, the optical sensor preferably functions to measure light signals (eg, images, intensities, etc.) from an individual's body region for arterial pressure data sets such as PPG data. . Optical sensors are preferably associated with user devices; however, optical sensors may be associated with parental devices, clinician devices, client devices, healthcare provider devices, and/or any suitable entity. . However, the optical device may be a separate device and/or component thereof and/or may have any suitable form.

[0039] The optical sensor may optionally include a light source. The light source may function to provide illumination (eg, uniform illumination, for low light conditions, for specific wavelengths of light, specific illumination intensity, etc.) for the collection of arterial pressure data sets. However, the light source may be ambient light (eg, sunlight, indoor lighting, etc.) and/or any suitable light source.

[0040] The data processing module 120 preferably functions to process the arterial pressure data set into a processed data set (e.g., cleaning such as outlier removal, noise removal, segmentation, filtering, etc.); However, the data processing module may process any suitable data set (eg, calibration data set, ECG data set, PPG data set, plethysmogram data set, etc.). The data processing module may perform one or more of the following processes on the dataset: normalization, background subtraction, smoothing, cleaning, transformation, fitting, interpolation, extrapolation, segmentation, decomposition (e.g. modal decomposition). ), noise removal, and/or any suitable step. The data processing module is preferably coupled to the data analysis module; however, the data processing module may communicate with the data analysis module and/or be arranged in any other suitable manner.

[0041] The data analysis module 130 preferably analyzes one or more data sets (eg, arterial pressure data sets, processed data sets, etc.) to determine one or more cardiovascular parameters of the individual. Function. The data analysis module may perform one or more of the following analyzes on the dataset: fitting (e.g., splines, curves, etc.), mathematical operations (e.g., scaling, differentiation, integration, transformations such as Fourier transforms, etc.), Root detection, correlation analysis, and/or any suitable analysis. Data analysis modules may include equations, lookup tables, conditionals, learning modules (eg, neural networks), and/or any suitable analysis tool.

[0042] The system optionally provides cardiovascular parameters, datasets (e.g., analyzed datasets, processed datasets, arterial pressure data sets, etc.), analysis (eg, baseline, cardiovascular manifold, diagnosis, etc.), and/or interface device 140 (eg, display) for presenting any suitable data or information. However, the data/information may be presented in any suitable manner.

[0043] However, system 100 may include any other suitable element configured to receive, process and/or analyze data to facilitate assessment or management of cardiovascular health of one or more individuals. can contain.

4. Method
[0044] Method 200 preferably functions to determine an individual's cardiovascular parameters based on an arterial pressure data set. The method is preferably performed by a system such as those described above, but may be performed by any system. One or more steps of the method can be performed sequentially (eg, serially) and/or in parallel (eg, simultaneously). Throughout the method, signals are preferably processed and analyzed in the time domain, but may alternatively be processed and analyzed in the frequency domain, in different domains at different steps, or in any other suitable domain. .

[0045] Measuring arterial pressure data sets S210 preferably collects data sets (eg, arterial pressure data sets, raw data sets, PPG data sets, plethysmogram data sets, etc.) in the individual's body region. function to S210 preferably occurs before processing the data set S220; however, S210 and S220 may occur simultaneously and/or at any suitable time. S210 is preferably performed by a data collection module (eg, the data collection module's optical sensors, strain gauges, etc.); however, any suitable component may be used. In some variations, one or more data sets may be stored (eg, in a database) and/or retrieved (eg, from a database). Data sets may be stored as raw data sets and/or as processed or analyzed data sets. The dataset is preferably measured at its predetermined frequency or range of 30-240 Hz (eg 60 Hz); however, at frequencies below 30 Hz, above 240 Hz, at variable frequencies and/or any can be measured at the appropriate frequency of Data sets are preferably measured over a predetermined length of time (e.g., 30 seconds, 45 seconds, 1 minute, 2 minutes, 5 minutes, 10 minutes, etc.); user's heart rate, personal activity status, etc.) and/or any suitable length of time.

[0046] S210 may optionally include receiving, obtaining and/or generating a supplemental data set. Examples of supplemental data sets include individual characteristics (e.g., height, weight, age, gender, race, ethnicity, etc.), individual (and/or individual family) medication history, individual activity level (e.g., , recent activities, past activities, etc.), medical concerns, medical professional data (eg, data from an individual medical professional), and/or any suitable supplemental data sets.

[0047] In certain examples, S210 includes positioning a data collection module (eg, on the individual's body region), positioning the individual's body region (eg, using a light source, using ambient light, etc.). Illuminating, measuring light scattering parameters (eg, reflection, absorption, etc.), and recording a series of light scattering parameters to generate an arterial pressure data set. Light scattering parameters are preferably measured at a data collection frequency of about 60 Hz or higher, although any suitable data collection frequency may be used. However, generating the arterial pressure data set may be performed in any suitable manner.

[0048] In a second particular example, S210 includes positioning a data collection module (eg, on a body region of the individual), determining the position of the individual (eg, using a light source, using ambient light, etc.). Illuminating the body region, measuring multiple images of the body region, and generating a photoplethymogram (PPG) data set from the multiple images. A PPG dataset may be generated from multiple images by extracting one or more features from the images, determining an optical flow depth map between images, recording changes in pixel properties across images, and/or It can be generated from multiple images in other ways.

[0049] In a third particular example, S210 may include one or more steps of data acquisition as disclosed in US patent application Ser. No. 16/538,361, which is incorporated herein in its entirety.

[0050] However, S210 may include any suitable steps and/or may be performed in any manner.

[0051] Processing the dataset S220 preferably converts the dataset (eg, raw dataset, arterial pressure dataset, etc.) to a processed dataset (eg, signal-to-noise ratio improvement, dataset segmentation, outlier removal from datasets, etc.). In a particular example, the raw data set contains (e.g., variation in contact pressure between the body region and the data acquisition module, variation in orientation between the body region and the data acquisition module, random noise, etc., during the measurement period). ) may contain a large non-stationary background. In this particular example, the non-stationary background can be removed by processing the dataset. However, any suitable data set may be generated and/or any suitable function may be performed by processing the arterial pressure data set.

[0052] Processing of the data set preferably occurs after measurement of the data set; however, processing of the data set may occur after any subset of the data set has been measured. S220 is preferably performed before analyzing the data set S230; however, S220 is contemporaneous with S230 and after S230 (e.g., processing data if an error occurs during data analysis). , and/or at any suitable time. S220 may preferably be performed by a data processing module; however, S220 may be performed by any suitable component. Data sets are preferably processed in a cloud computing system, but may be processed on user devices, local computing systems and/or any suitable location.

[0053] For example, as shown in FIGS. 3 and 4, S220 includes resampling the data set S221; filtering the data set S223; segmenting the data S225; determining a subset of data to analyze S228; and/or any suitable step.

[0054] S221 preferably inserts new data points into the data set (eg, raw data set, arterial pressure data set, PPG data set, plethysmogram data set, etc.) and/or (eg, existing function to remove irrelevant data points (between data points) so that datasets (interpolated datasets, interpolated raw datasets, etc.) are placed at intervals of pre-determined frequency/duration contains data points that are S221 is preferably performed before S226; however, S221 may be performed at the same time as S226 and/or after S226. The default frequencies are preferably multiples of 60 Hz (e.g., 120 Hz, 240 Hz, etc.); however, the default frequencies are selected based on the number of data points in the data set (e.g., total number of data points). can vary (e.g., low in data-sparse regions, such as for sampling datasets with a small number of data points; data-rich regions, such as for sampling datasets with a large number of data points) ) and/or any suitable frequency may be used. New data points are preferably determined by interpolating between nearest neighboring data points, but additionally or alternatively, machine learning can be used by interpolating between any suitable neighboring data points. by fitting the data to a curve, by using simulations, and/or by generating additional data in other ways. Interpolation is preferably linear; however, non-linear interpolation and/or any suitable interpolation may be used. Data points are preferably removed such that the time interval between data points is constant (e.g. based on a predetermined frequency), but additionally or alternatively bad data (e.g. outliers, artifacts, etc.) and/or may be deleted in other ways.

[0055] S223 preferably extracts the data set (e.g., arterial pressure data set, raw data set, interpolated data set, interpolated data set) by applying one or more filters to generate a filtered data set. It functions to improve the signal-to-noise ratio (SNR) of raw data sets, etc.). S223 is preferably performed before S225; however, S223 may be performed at the same time as S225 and/or after S225. The filter is a short-pass filter (e.g., with cutoff frequencies below 0.1 Hz, 0.1, 0.5, 1, 10, 20, 30, 60, 100, 200, 500, 1000, above 1000 Hz, etc.) , longpass filters (e.g., with cutoff frequencies below 0.1 Hz, 0.1, 0.5, 1, 10, 20, 30, 60, 100, 200, 500, 1000, above 1000 Hz, etc.), bandpass It may be a filter (eg, by combining a longpass filter and a shortpass filter), a notch or bandstop filter, and/or any suitable filter. In certain examples, the filter may be a longpass filter capable of removing signals below 0.5 Hz from the data set; however, any suitable cutoff filter may be used. Filters may be applied to the dataset in the time domain and/or the frequency domain. The entire dataset is preferably filtered simultaneously; however, one or more subsets of the dataset may be filtered at a time and/or filtering of the dataset may be performed in any suitable manner. In the illustrated example, each segment of the segmented data set (eg, the segmented data set as generated in steps S225, S226 or S228) is filtered (eg, using the same or different filters) can be In a second illustrative example, the raw dataset and/or the interpolated dataset may be filtered. However, any suitable data set or sets may be filtered.

[0056] S225 preferably includes each separate physiological cycle (eg, heartbeat, cardiac cycle, respiratory cycle, digestive cycle, etc.) and data set (eg, arterial pressure data set, interpolated data set, filtered data set) and to generate a segmented data set. The segmented data set is preferably segmented into individual beats; however, the segmented data set is a set of beats (e.g. beats with little change in arterial pressure waveform, can be segmented into individual arterial pressure waveform components (e.g. direct transmission; reflected arterial pressure signals from iliac arteries, renal arteries, etc.); can be segmented into an appropriate form of S225 preferably occurs before S226; however, S225 may occur concurrently with and/or after S226. Each segment is preferably the same size (eg, same time window, same frequency window, etc.); however, the segments may have any suitable size. Segments are preferably distinct (eg, non-overlapping); however, segments may overlap and/or have any suitable relationship with other segments.

[0057] The segmented data set is preferably segmented using slow and fast moving averages (e.g., moving average crossover); method, threshold method, variational method, machine learning (e.g., using a trained neural network), pattern matching (e.g., segmented whenever a dip is detected), and/or may be segmented in any suitable manner. In a particular example, beat segmentation using slow moving averages and fast moving averages is applied to each point of the interpolated data set.

[0058] S226 preferably denoises the dataset (eg, raw dataset, interpolated dataset, filtered dataset, segmented dataset, etc.) to improve the SNR of the dataset. , and functions to generate a denoised data set (eg, a denoised segmented data set). S226 preferably occurs before S228; however, S226 may occur concurrently with and/or after S228. S226 is preferably performed after S225, but may be performed before and/or concurrently with S225.

[0059] In certain variations, S226 may include decomposing the data set into eigenmodes. These variations may speed up processing, reduce processing resources, and/or perform any suitable function by decomposing the data into eigenmodes. Each segment of the segmented data set is preferably independently decomposed into eigenmodes. However, the dataset (e.g., segmented dataset, filtered dataset, raw dataset, etc.) may be decomposed into eigenmodes as a whole, multiple segments may be decomposed together, and/or The dataset may be decomposed into eigenmodes in any suitable manner. Eigenmodes may correspond to discrete modes and/or continuous modes (eg, frequencies of continuous transformation, etc.).

[0060] In a particular example of this particular variation, each eigenmode may encode different capacities of signal data (eg, data relating to cardiovascular parameters) and noise data. Encoding signal data (e.g., primarily 1st, 2nd, 3rd, 4th, 5th, 6th, 7th, 8th, 9th, 10th, 15th, 20th, etc. signal data) The specific eigenmodes that encode the data) may be included in the denoised data and the eigenmodes that encode the noise data (e.g., mainly 1st, 2nd, 3rd, 4th, 5th , any mode after the 6th, 7th, 8th, 9th, 10th, 15th, 20th, etc.) modes are excluded from the denoised data. be. Including the modes in the denoised data set preferably includes adding each of the modes for inclusion in the data set, but additionally or alternatively calculating the average of the modes to be included. , and/or combining the eigenmodes in other ways. The eigenmodes that mainly encode the data are the first mode, the first two modes, the first three modes, the first four modes, the first five modes, the first ten modes, the second to fourth mode, second to sixth mode, first half mode, last half mode, first third mode, second third mode, last third mode , the first quarter mode, the first tenth mode, a subset thereof, and/or any suitable mode. However, denoised data may be generated in any suitable manner.

[0061] The eigenmodes are preferably determined using extended empirical mode decomposition (EEMD); Analysis (PCA), any suitable basis set for the dataset (orthogonal basis set, orthonormal basis set, etc.), fitting the dataset (e.g., polynomial fit, noise model, to equations, etc.), and/or any can be determined using any suitable method of Eigenmodes may be determined in the frequency domain and/or the time domain. Eigenmodes are preferably determined separately for each segment (e.g., arterial pressure data associated with a given heartbeat); however, fixed mode analysis may be performed on the entire dataset, and /or any suitable subset of the dataset may be used.

[0062] In certain examples, EEMD analysis of a dataset may decompose the dataset into about 20 eigenmodes (eg, ±5 modes, ±10 modes, ±20 modes, etc.). In this example, eigenmodes 1-6 may be selected to generate the denoised data set; however, any suitable mode may be used. A denoised data set may be generated by summing eigenmodes 1-6 (eg, excluding the remaining eigenmodes). However, denoising the data set may be performed in any suitable manner.

[0063] S228 preferably extracts one or more subsets of the data set (e.g., denoised data set, segmented data set, filtered data set, interpolated data set, raw data set, etc.) to determine and contain the processed dataset. A subset of the dataset is determined based on the SNR within the subset; however, based on data completeness, moderate accuracy and/or any suitable data quality metric.

[0064] The processed dataset preferably comprises any suitable number and/or range of segments between 1 and 1000, such as 30 segments; , segmented dataset, filtered dataset, interpolated dataset, raw dataset, etc.) may be included in the processed dataset and/or any suitable number of segments may be included in the can be included. The segments included in the processed dataset are preferably continuous (e.g., do not skip any beats); however, the processed dataset may skip any number of suitable segments, and /or may include any suitable data. The number of segments included in the data set may depend on the calibration method, the cardiovascular parameter to be determined, the target and/or threshold duration, and/or may be determined in other ways. In the illustrated example, if the cardiovascular parameter to determine includes blood pressure, the processed data set preferably includes about 10 consecutive segments (e.g., about 10 discrete cardiac cycles, such as consecutive heart beats). detectable). However, any number of consecutive segments can be used.

[0065] In one particular embodiment of S228, the subset of the dataset may be determined based on the moving time window SNR. The time window may be any suitable time window or range thereof from 0 to 120 seconds, such as 15 seconds; however, any time window may be used. The time window preferably begins with one segment (eg, one heartbeat) of the data set, but may additionally or alternatively encompass multiple heartbeats or portions thereof. After comparing the SNR within the time window to the SNR threshold (e.g., 1, 2, 10, 20, etc.), if the SNR within the time window is greater than or equal to the SNR threshold, the segment within the time window is the processed data set and subsequent time windows may be examined (eg, the time window to be examined may be shifted by one or more segments and the analysis repeated). However, if the SNR within the time window is less than the SNR threshold, the segment may be rejected. If a suitable subset of the dataset cannot be determined (e.g., many segments are rejected), a new dataset can be measured (e.g., by repeating S210) (e.g., steps S221-S226, etc.). The data set may be reprocessed (by repeating one or more steps of S220 of ) and/or any suitable process may be performed. However, the segments included in the processed dataset may be determined in any suitable manner.

[0066] Analyzing the data set S230 preferably includes the data set (eg, processed data set, denoised data set, segmented data set, filtered data set, interpolated data set, set, raw data set, etc.) to determine one or more cardiovascular parameters of the individual. S230 may additionally or alternatively function to determine criteria related to an individual's cardiovascular parameters (and/or any other suitable parameters). The dataset is associated with an averaged dataset (e.g., the subset of the dataset identified at S228) for the entire dataset, segment by segment (e.g., cardiovascular parameters are determined for each segment). averaged segments with other segments on a per time step basis) and/or analyzed in other ways. S230 is preferably performed by a data analysis module; however, any suitable component may be used. S230 is preferably performed on a cloud computing system, but may be performed on a user device, on a local computing system and/or by any computing system. S230 is preferably performed independently for each segment of the data set; although S230 may be performed for the entire data set, the analysis of one segment may depend on the results of other segments. , and/or any suitable subset of the data set may be analyzed. The analysis is preferably transmitted to a user device and/or interface device, but may additionally or alternatively be transmitted to a healthcare provider, parent, database and/or any suitable endpoint.

[0067] Cardiovascular parameters may be evaluated using datasets, criteria, and/or regressive modeling (eg, linear regression, nonlinear regression, etc.), learning (eg, trained neural networks, machine learning algorithms, etc.). It may be determined based on vessel manifolds, equations, lookup tables, conditional statements, transformations (eg, linear transformations, non-linear transformations, etc.) and/or in any suitable manner.

[0068] Transformations (eg, correlations) between baseline and/or cardiovascular variables and cardiovascular parameters are preferably performed using calibration datasets (eg, calibration datasets such as from blood pressure cuffs, analysis datasets and Such as ECG measurements generated at approximately the same time; Second arterial pressure data sets such as individuals in different body regions, different individuals, individuals with different activity states; Analysis of each individual in a control group with corresponding measured cardiovascular parameters. data set, including a calibration data set); however, the transformation is based on a model (e.g., a model of the individual cardiovascular system, a global model such as a model applicable to any user, etc.) may be determined and/or determined in any suitable manner.

[0069] S230 may include determining a baseline S232, determining a cardiovascular parameter S236, and storing the cardiovascular parameter S239. However, S230 may include any suitable process.

[0070] S232 preferably processes data sets (e.g., processed data sets, denoised data sets, segmented data sets, filtered data sets, interpolated data sets, raw data sets, etc.) ) to determine the criteria for S232 preferably occurs before S236; however, S232 may occur concurrently with and/or after S236. The set of criteria may depend on cardiovascular parameters, individual characteristics, supplemental data sets and/or any suitable information. In some variations, different criteria may be used for different cardiovascular parameters; however, two or more cardiovascular parameters may be determined from the same set of criteria.

[0071] In a first variation of S232, determining the criterion optionally comprises fitting a spline to the processed data set, a derivative of the spline fit data set (e.g., first order, 2nd order, 3rd order, higher order, etc.) and determining roots (eg, zeros) in one or more of the derivatives. The criteria may be one or more of the root of the derivative, the value of the dataset and/or other derivatives evaluated at the root of the derivative and/or may be otherwise determined. In a particular first example, the criteria may include two roots of the second derivative and one root of the third derivative; evaluated), the amplitude of any suitable derivative of the dataset (e.g., evaluated at one of the roots), and/or the value of any suitable root and/or dataset. In a particular second example, the criteria are the values of the processed data set and the first four zeros of the first and second derivatives of the processed data set (e.g., relative to the start of the segment, the may correspond to the first and second derivatives of the processed data set evaluated at corresponding time points to the peak of the processed data, to the previous cardiac cycle, etc.). In a particular third example, as shown in FIG. 8, the criteria are the amplitude of the processed data set (and/or one or more segments thereof), the amplitude of the processed data set (and/or one or more segments thereof), segment), the amplitude of the second derivative of the processed data set (and/or one or more segments thereof), and the first derivative, the second derivative and/or the third It may correspond to a set of amplitudes of the third derivative of the processed data set (and/or one or more segments thereof) evaluated at the root of the derivative and/or any subset thereof. In FIG. 8, the zero of each derivative is identified as a point in time of interest (illustrated by a black circle in FIG. 8), and the value of each derivative at each point in time of interest is a reference candidate. In a particular fourth example, the criteria may be selected from and/or include any of the following values.
[0072] p(g), p''(g), p''(g), p(h), p'(h), p'''(h), p(k), p'( k), p''(k)
[0073] Where p(a) is the processed data set (and/or a segment thereof) at time a, p' is the first derivative of p with respect to time, and p'' is p with respect to time. is the second derivative p of p, p''' is the third derivative of p with respect to time, g corresponds to the sequence of times for which p'(g)=0, and h is p'' Corresponds to one or more times at which (h)=0, and k corresponds to one or more times at which p'''(k)=0.

[0074] In a second variation of S232, determining the criteria may include decomposing the processed data set (eg, for each segment of the analysis data set) into any suitable basis functions. In particular examples, decomposing the processed data set includes discrete Fourier transforms, fast Fourier transforms, discrete cosine transforms, Hankel transforms, polynomial decompositions, Rayleighs, wavelets and/or any suitable transforms for the processed data set. It may include performing a decomposition and/or transformation. The criteria may be one or more of decomposition weights, phases and/or any suitable output of the decomposition. However, the criteria can be determined in any suitable way from the data set.

[0075] In a third variation of S232, determining the criteria may include fitting the processed data set to a predetermined functional form. Functional forms may include Gaussian, Lorentzian, exponential, supergaussian, Lévy, hyperbolic secant, polynomial, convolution, linear and/or nonlinear combinations of functions, and/or any suitable function. The fit can be constrained or unconstrained. In certain examples, a linear combination of five constrained Gaussian distributions (eg, based on the user's cardiovascular status and/or phase) may be used to fit each segment of data. However, any suitable fitting may be performed. Criteria are preferably one or more of the fit parameters (e.g., full width at half maximum (FWHM), center position, amplitude, frequency, etc.); and/or may include any suitable information.

[0076] Determining cardiovascular parameters S236 preferably functions to determine the cardiovascular status (eg, set of cardiovascular parameter values) of the user. Cardiovascular parameters may be determined based on criteria, based on cardiovascular variables, and/or determined in other ways. S236 preferably determines a cardiovascular parameter for each segment of the dataset (e.g., each heart beat); ), a single cardiovascular parameter, and/or any suitable information. S236 is preferably performed before S239; however, S236 may be performed at the same time as S239 and/or after S239.

[0077] In a first variation of S236, cardiovascular parameters may be determined by applying a criterion transformation to a set of criteria (eg, determined at S232), eg, as shown in FIG. 7B. Reference transformations can be determined from the calibration data set based on a model (e.g., a model of the individual, a model of human anatomy, a physical model, etc.) (e.g., a set of different individual reference transformations, as shown in FIG. 7A). are determined by multiplying the cardiovascular parameters by the inverse of the respective criterion), generated using machine learning (e.g., neural networks), fits (e.g., least-squares fit, non-linear least-squares fit, etc.) ) (eg, associating reference value sets with cardiovascular parameter values) and/or otherwise determined. A reference transformation can be a common transformation, can be unique to a given cardiovascular parameter or a combination thereof, can be specific to an individual parameter (e.g., age, demographics, comorbidities, biomarkers, medications, estimated or measured physiological state, etc.), may be individual-specific, may be specific to the measurement context (eg, time of day, ambient temperature, etc.), or may be otherwise generic or specific. Reference transformations can be based on a subset of controls (e.g., a subset of controls with one or more characteristics similar to or consistent with the individual's characteristics), based on the mean, median, most accurate (e.g., smallest residual, smallest error, etc.), can be selected based on voting, can be selected by a neural network, can be randomly selected, and/or can be otherwise determined from a calibration data set.

[0078] The reference transformation (eg, linear regression coefficients) can be a linear transformation or a non-linear transformation. Each cardiovascular parameter may be associated with a different reference transformation and/or one or more cardiovascular parameters may be associated with the same reference transformation (e.g., two or more cardiovascular parameters may be correlated or covariate can be). In a first variant of certain examples, cardiovascular parameters may be determined according to
[0079] AT=B
[0080] A corresponds to a set of criteria, T corresponds to a criteria transformation, and B corresponds to a cardiovascular parameter.

[0081] In certain examples, the method includes determining a baseline transformation for the individual and determining a cardiovascular parameter value for the individual based on the subsequent cardiovascular measurements and the baseline transformation. The reference transformation is preferably determined from a set of calibration data sampled from the individual, the set of calibration data being reference (e.g. PPG data, plethysmogram data) extracted from calibrated cardiovascular measurements (A). , and calibrated cardiovascular parameter measurements (eg, blood pressure, _O2 level, etc.; measurements of cardiovascular parameters to be determined) (B). The individual canonical transform (T) is determined from AT=B. T is then used to determine a baseline cardiovascular parameter value extracted from subsequently sampled cardiovascular measurements.

[0082] In a second variation of S236, cardiovascular parameters are determined from where the individual is on the individual's cardiovascular manifold, from the individual's cardiovascular manifold to the common cardiovascular manifold, for example, as shown in FIG. 6B. It may be determined based on a manifold transformation to a body and, optionally, a mapping transformation from individual locations on a common cardiovascular manifold to cardiovascular parameter values. Cardiovascular parameters may additionally or alternatively be a change in where an individual is on a cardiovascular manifold, an individual's effective position on a common cardiovascular manifold, an individual's effective position on a common cardiovascular manifold. and/or otherwise dependent on the individual's relationship to the cardiovascular variant. Common cardiovascular manifolds can be determined from a calibration dataset (e.g., shown in FIG. 6A), can be determined from a model, can be generated using machine learning (e.g., neural networks), and/or It can be determined in other ways. The common cardiovascular manifold can be the mean of the calibration data set, can include extreme values of the calibration data set, can be learned from the calibration data set (e.g., determined using a machine learning algorithm), calibration data may be selected from the set and/or otherwise determined based on the calibration data set. A common cardiovascular manifold preferably maps values of one or more criteria to values of cardiovascular parameters, but may be constructed in other ways. A common cardiovascular variant preferably encompasses at least a majority of the possible baseline and/or cardiovascular parameter values of the population, but may encompass other suitable swaths of the population. . The common cardiovascular manifold may be specific to one or more cardiovascular parameters (eg, the system may include different common manifolds for blood pressure and oxygen levels), but alternatively multiple or all of the subjects of cardiovascular parameters. Manifold transformations may include one or more affine transformations (eg, any combination of one or more of transformations, scaling, similarity, similarity transformations, reflection, rotation and shear mappings) and/or any suitable transformations. . In an illustrative example of the second variant, an individual's cardiovascular phase is determined, and (e.g., using a transform) the individual's cardiovascular phase to a common cardiovascular phase (e.g., a common cardiovascular diversity The relationship between common cardiovascular phases and cardiovascular parameters is known.

[0083] In certain examples, the method includes generating a common manifold from the population calibration data, generating an individual manifold from the individual calibration data, and generating a common manifold with the individual manifold. Involves determining transformations to and from manifolds. The common manifold is preferably a finite area and encompasses all (or most) perturbations and corresponding cardiovascular parameter values (e.g., responses), but any other suitable space. can be included. A common variant preferably associates a combination of criteria (with different values) with values of various cardiovascular parameters, but may associate other variables. Personal calibration data preferably includes cardiovascular measurements (eg, PPG data, plethysmogram data) corresponding to cardiovascular parameter measurements (eg, blood pressure), but may include other data. Population calibration data preferably includes data similar to individual calibration data, but across multiple individuals (eg, in one or more physiological states). Transformations may be calculated (eg, as equations, as constants, as matrices, etc.), estimated, or otherwise determined. The transformation preferably represents a transformation between an individual manifold and a common manifold, but additionally or alternatively a reference position on the common manifold (e.g. a specific reference value transformed to a common region ) to cardiovascular parameter values (eg, within a common region). Alternatively, the method may apply a second transformation to transform the universally transformed reference values into cardiovascular parameter values (eg, in common regions). The transform is then applied to criteria extracted from subsequent cardiovascular measurements from the individual to determine the individual's cardiovascular parameter value.

[0084] Embodiments of S236 may include determining a cardiovascular variant of the individual. For example, as shown in FIG. 5, an individual's cardiovascular manifolds may correspond to surfaces related to changes in an individual's cardiac function, nervous system, and blood vessels. However, cardiovascular variants may additionally or alternatively depend on the endocrine system, immune system, digestive system, renal system and/or any suitable system of the body of an individual. A cardiovascular manifold may additionally or alternatively be a volume, a line, and/or otherwise represented by any suitable shape. An individual's cardiovascular manifold is preferably substantially constant (eg, slowly changing day-to-day, week-to-week, month-to-month, year-to-year, etc.) over the life of the individual. Thus, an individual's cardiovascular variants can be stored, accessed, and used for later analysis of an individual's cardiovascular parameters. However, an individual's cardiovascular variants may be variable and/or may change significantly (e.g., as a result of significant blood loss as a side effect of drug therapy, etc.) and/or any It may have other properties.

[0085] In some variations, the cardiovascular manifold may correspond to and/or be derived from a predefined functional form (eg, from the third variant of S232). However, cardiovascular variants may and/or not be associated with criteria in other ways.

[0086] The cardiovascular manifold preferably corresponds to a hyperplane, but additionally or alternatively to a triangular manifold, a sigmoidal manifold, a hypersurface, a higher-order manifold and/or any suitable can be described by a phase space

[0089]

はデータセットのセグメント、ｔは時間、Ｎはフィットである関数の総数、ｉはフィットの各関数のインデックスであり；ａ、ｂ及びｃはフィットパラメータであり、ｐ_ｘｉは心血管位相＜φ＞の関数であり、フィットパラメータはｐ_ｘｉの値に拘束される。拘束関数は、フィットパラメータごとに同じであり得る又は異なり得る。拘束関数は、好ましくは、連続的に微分可能であるが、既定の時間ウィンドウにわたって連続的に微分可能であり得、及び／又は、連続的に微分可能ではない。拘束関数の例には、定数、線形項、多項式関数、三角関数、指数関数、ラジカル関数、有理関数、それらの組み合わせ及び／又は任意の適切な関数が含まれる。 [0087] As shown for example in FIG. 9, determining an individual's cardiovascular manifold may involve multiple segments of a dataset (eg, a segmented dataset, a processed dataset, a subset of the dataset, etc.). ) to multiple Gaussian functions such as
[0088]

[0089]

_is the segment of the data set, t is the time, N is the total number of functions that are fits, i is the index of each function in the fit; and the fit parameters are constrained to the values of p _xi . The constraint function can be the same or different for each fit parameter. The constraint function is preferably continuously differentiable, but may be continuously differentiable over a predetermined time window and/or not continuously differentiable. Examples of constraint functions include constants, linear terms, polynomial functions, trigonometric functions, exponential functions, radical functions, rational functions, combinations thereof and/or any suitable function.

[0090] In a third variation, determining the cardiovascular parameter may comprise determining the cardiovascular parameter based on supplemental data. For example, the canonical and/or manifold transformations may be modified based on supplemental data (such as to account for known biases or offsets regarding an individual's gender or race).

[0091] In a fourth variation, cardiovascular parameters may be determined in a number of ways. For example, cardiovascular parameters may be determined according to two or more of the above variations. In a fourth variant, the individual cardiovascular parameter is the average cardiovascular parameter, the most probable cardiovascular parameter selected based on voting, the most extreme cardiovascular parameter (e.g. highest, lowest, etc.). obtained, dependent on previously determined cardiovascular parameters, and/or selected in other ways.

[0092] However, cardiovascular parameters may be determined in any suitable manner.

[0093] The method may optionally include determining a classification of the cardiovascular condition. Examples of cardiovascular status classes (CSC) include rest, exercise, deep breathing, shallow breathing, sympathetic activation, parasympathetic activation, hypoxia, hyperoxia and/or other categories. The CSC is preferably determined based on the values of a set of cardiovascular parameters, but additionally or alternatively, ancillary data (e.g. contextual data such as time of day, ambient temperature, medications taken; physiological measurements, etc.). ), and/or other data. CSCs include classifiers (e.g., RNN, CNN, autoencoders, KNN, etc.), correlations, rules, heuristics (e.g., if a set of predefined cardiovascular parameters have values within their respective range sets, the CSC ) and/or may be determined in other ways.

[0094] In an embodiment, S236 may include predicting the effect of one or more predetermined activities on the individual's cardiovascular parameters. A predicted effect is defined as a set of predicted criteria for when an individual performs a given activity, a predicted location on the individual's cardiovascular manifold, a predicted effect on a common cardiovascular manifold, position, can be determined based on a simulated data set (eg, a raw data set, a processed data set, a simulated data set similar to the data set generated by steps S210-S228, etc.), and/ Or it can be predicted in other ways.

[0095] S239 preferably functions to store cardiovascular parameters (and/or any suitable data sets) in client databases, clinician databases, user databases and/or suitable locations/components. . S239 optionally tracks cardiovascular parameters over time (e.g., 1 hour, 1 day, 1 week, 1 month, 1 year, 10 years and/or any time therebetween). can include For example, an individual's cardiovascular parameters may be tracked over a predetermined period of time to determine stability to cardiovascular parameters, how engagement in activity (e.g., predetermined activity) affects cardiovascular parameters, and/or other Individual cardiovascular parameters can be tracked in the following manner. However, S239 can be used in other ways.

[0096] S240 preferably presents the cardiovascular parameter (and/or any suitable data set or analysis thereof) to an individual, clinician, healthcare provider, parent and/or any suitable individual. function to (e.g., display). S240 is preferably performed on an interface device (e.g., a user device, a healthcare provider device, a parent device, a clinician device, etc.); however, any suitable component may be used to perform the analysis. can present. Cardiovascular parameters are visually displayed numerically (e.g., as percentiles, as absolute values, etc.) in tables using indicators (e.g., "good" or "worse") as changes relative to previous measurements may be displayed and/or displayed in other ways. S240 may occur before, during and/or after S230.

[0097] In variations, S240 may present one or more predefined activities to the individual (and/or the clinician, healthcare provider, parent, etc.). In these variations, the predetermined activity is preferably selected to induce a positive change in the user's cardiovascular parameter, but may additionally or alternatively present a negative change (e.g., if the individual attempts to discourage participation in routine activities) and/or no change. The default action may be selected based on the expected impact of the default action (eg, as predicted in variations of S236), the probability that the default action will have an impact.

[0098] In a related variation, S240 may include presenting an analysis of the individual's cardiovascular parameters over time, which shows how one or more activities involving the individual affect the individual's cardiovascular parameters. can indicate whether it affects The time period may be minutes, hours, days, weeks, months, years, decades, and/or any suitable time period therebetween to provide long-term or May exhibit short-term (eg, immediate) effects.

[0099] S250 may optionally include determining a physiological state that may function to determine a value for a set of physiological parameters. Examples of physiological parameters that can be determined include cardiovascular parameters, body temperature, hormonal parameters (eg, levels, rates of change, etc.), immunological function and/or other parameters. Physiological status may be determined based on cardiovascular parameter values, reference values, auxiliary information (eg, time of day, ambient temperature, dosing schedule, etc.), and/or other information. Physiological state may be determined over a predetermined time window, multiple time frames (e.g., physiological state may be determined based on parameter changes over time), and/or information about other temporal data sets. can be determined based on Physiological states may be determined using neural networks (e.g., trained based on historical data), correlations, lookup tables, secondary common manifolds that map inputs to physiological states. , or may be determined in other ways.

[00100] In a particular example, method 200 includes: measuring a raw PPG dataset (eg, at a frame rate of 60 Hz); interpolating the raw PPG dataset to a standard frame rate (eg, 240 Hz); generating an interpolated raw data set; generating a filtered data set by filtering the interpolated raw data set (e.g., a long pass filter to remove signals below 0.5 Hz); generating a segmented raw data set by segmenting the interpolated raw data set (e.g., based on slow and fast moving averages, into arterial pressure waveforms corresponding to individual heartbeats, etc.); Denoising generating a segmented data set; based on the signal-to-noise ratio of the moving time window of the denoised segmented data set, the processed data set (e.g., the denoised segmented determining criteria from the processed data set (e.g., criteria for each segment of the analysis data set); and determining an individual's cardiovascular parameters based on the criteria. obtain. In this particular example, determining the criterion from the analytical data set includes fitting the analytical data set to a spline function; calculating first, second and third derivatives of the fitted analytical data set; It may include determining the roots of the second, second and third derivatives. However, cardiovascular parameters may be determined in any suitable manner.

[00101] System and/or method embodiments may include all combinations and permutations of various system components and various method processes, and one or more examples of the methods and/or processes described herein may include: , asynchronously (eg, serially), concurrently (eg, in parallel), or by and/or used by one or more instances of the systems, elements and/or entities described herein. and in any other suitable order.

[00102] Those of ordinary skill in the art will recognize from the foregoing detailed description and drawings, and from the claims, that the preferred embodiments of the invention can be implemented without departing from the scope of the invention, which is defined in the following claims. Modifications and changes may be made.

Claims (21)

A method for determining the blood pressure of a patient, comprising:
receiving a plurality of images of a body region of a user, said images being recorded using an image recording device of a user device associated with said patient, said image recording device said body region of said user; receiving in contact with
generating a photoplethymogram (PPG) dataset from the plurality of images;
filtering the PPG data set using a low pass filter configured to remove signals below 0.5 Hz from the PPG data set;
decomposing the PPG data set into a set of modes using an extended empirical mode decomposition;
extracting a fixed mode from the set of modes;
denoising the PPG data set to generate a denoised PPG data set by aggregating the fixed modes to generate a denoised PPG data set;
segmenting the denoised PPG data set into a segmented PPG data set using moving average segmentation;
determining a processed PPG dataset by averaging a predetermined number of segments of the segmented PPG dataset;
extracting a set of criteria from the processed PPG dataset;
determining a patient's cardiovascular manifold location based on the set of criteria;
transforming the patient's cardiovascular manifold location to a common cardiovascular manifold location;
determining the blood pressure of the patient based on the mapping of locations on the common cardiovascular manifold to blood pressure values. 2. The method of claim 1, wherein the predetermined number of segments comprises at least ten segments of the PPG dataset. at least one of said common cardiovascular manifolds or said transformations,
For each patient in the group of calibrated patients,
measuring blood pressure using at least one of a sphygmomanometer, a digital blood pressure monitor, a radial artery tonometry, an arterial catheter or an electrocardiogram;
determining a set of criteria corresponding to the calibration patient concurrently with measuring blood pressure;
determining a relationship between the measured blood pressure and the set of criteria;
and combining the relationships for each patient to determine at least one of the common cardiovascular manifold or the transformation. Extracting the set of criteria from the processed PPG data set includes fitting the processed PPG data set to a fitting function, wherein the set of criteria defines one or more fit parameters of the fitting function. 2. The method of claim 1, comprising: 5. The method of claim 4, wherein the fitting function comprises multiple Gaussian distributions. 5. The method of claim 4, wherein the fit parameters are constrained to hyperplane values. A method for determining the cardiovascular status of a patient, comprising:
receiving a plethysmogram (PG) dataset associated with the patient;
removing noise from the PG dataset by summarizing a set of fixed modes corresponding to the PG dataset;
segmenting the denoised PG data set into segmented PG data sets, each segment corresponding to a cardiac cycle of the patient;
extracting a set of criteria for the segmented PG dataset, the set of criteria comprising:
the segmented PG dataset;
a first derivative of the segmented PG dataset;
determined based on the value of at least one of the second derivative of the segmented PG data set or the third derivative of the segmented PG data set;
determining the patient's cardiovascular status by applying a transformation to the set of criteria. 8. The method of claim 7, further comprising interpolating between data points of the PG data set to fill gaps in the PG data set. 8. The method of claim 7, wherein the patient's cardiovascular status comprises one or more of the patient's blood pressure, cardiac events, cardiovascular drift, and drug response. 8. The method of claim 7, wherein extracting a set of criteria comprises extracting a set of criteria for each segment of the segmented PG dataset. Extracting the set of criteria for the segment comprises:
calculating the first, second and third derivatives of the segment;
determining the time points corresponding to zeros of the first, second and third derivatives;
determining the value of the segment at the time point;
and determining values of the first, second and third derivatives of the segment at the time point. The set of criteria includes:
values of the segmented PG data set at time points corresponding to the first four zeros of the first and second derivatives;
values of the first derivative of the processed PG data set at times corresponding to the first four zeros of the first and second derivatives;
and values of the second derivative of the processed PG data set at times corresponding to the first four zeros of the first and second derivatives. 8. The method of claim 7, wherein reducing noise in the PG data set further comprises filtering the PG data set using a long pass filter. 8. The method of claim 7, wherein the set of fixed modes is determined using extended empirical mode decomposition. 15. The method of claim 14, wherein denoising from the PG data set comprises summing the first six modes of the set of fixed modes. 8. The method of claim 7, wherein segmenting the denoised PG data set comprises segmenting the denoised PG data set by detecting moving average crossovers. further comprising determining an average PG data set by averaging a predetermined number of segments of the segmented data, wherein the reference set is determined based on the averaged PG data set. 8. The method of claim 7. determining a cardiovascular variant of the patient based on the set of criteria;
determining a manifold transformation that transforms the patient's cardiovascular manifolds into a common cardiovascular manifold;
8. The method of claim 7, wherein said transformation comprises said manifold transformation. 8. The method of claim 7, wherein the transform is determined based on a calibration data set generated based on cardiovascular status measurements derived from a group of calibration patients. 8. The method of claim 7, wherein said transform is a linear transform. A system configured to perform the method disclosed in any one of claims 1-20.

Images